基于支持向量机的分类预测算法研究

Research Classification Algorithm Based on Support Vector Machine

分类预测是数据挖掘、机器学习和模式识别等很多领域共同关注的问题,已经存在了许多有效的分类算法,但这些算法还不能解决所有的问题。支持向量机作为一种新的分类预测工具,能根据有限样本信息在模型的复杂性和学习能力间取得平衡,并能获得更好的泛化能力。SMO算法是支持向量机中使用最多的算法,它体现了支持向量机的优点,同时也能处理大规模训练集。

Classification is a common concern in many areas, such as data mining, machine learning, pattern recogmnon, etc. There have been many effective classification algorithms. However, these algorithms still can not solve all the problems. Support Vector Machine achieves a balance between complexity of the model and ability to learn according to finite sample information. And it can get better generalization ability. SMO algorithm is most commonly used algorithm in SVM. It embodies the advantages of SVM. It is also capable of handling large-scale training set at the same time.